A lightweight explicit alignment recipe that adapts off-the-shelf VLMs into robust omni-modal embedding models. https://arxiv.org/abs/2601.03666
Haonan Chen PRO
Haon-Chen
AI & ML interests
None yet
Recent Activity
new activity
5 days ago
AdaTooler-V/AdaTooler-V-train-data:Upload json file
upvoted
a
paper
7 days ago
e5-omni: Explicit Cross-modal Alignment for Omni-modal Embeddings
updated
a collection
7 days ago
e5-omni
Organizations
Aligned embedding data synthesis models and embedding model. Our paper: https://arxiv.org/pdf/2410.18634
HomePage: https://haon-chen.github.io/MoCa/
mmE5: Improving Multimodal Multilingual Embeddings via High-quality Synthetic Data
-
intfloat/mmE5-mllama-11b-instruct
Zero-Shot Image Classification • 11B • Updated • 129 • 20 -
mmE5: Improving Multimodal Multilingual Embeddings via High-quality Synthetic Data
Paper • 2502.08468 • Published • 16 -
intfloat/mmE5-synthetic
Viewer • Updated • 560k • 461 • 6 -
intfloat/mmE5-MMEB-hardneg
Viewer • Updated • 1.47M • 569 • 1
A lightweight explicit alignment recipe that adapts off-the-shelf VLMs into robust omni-modal embedding models. https://arxiv.org/abs/2601.03666
HomePage: https://haon-chen.github.io/MoCa/
mmE5: Improving Multimodal Multilingual Embeddings via High-quality Synthetic Data
-
intfloat/mmE5-mllama-11b-instruct
Zero-Shot Image Classification • 11B • Updated • 129 • 20 -
mmE5: Improving Multimodal Multilingual Embeddings via High-quality Synthetic Data
Paper • 2502.08468 • Published • 16 -
intfloat/mmE5-synthetic
Viewer • Updated • 560k • 461 • 6 -
intfloat/mmE5-MMEB-hardneg
Viewer • Updated • 1.47M • 569 • 1
Aligned embedding data synthesis models and embedding model. Our paper: https://arxiv.org/pdf/2410.18634